Solid electrolytes and electrical interconnects for oxygen delivery devices

Solid State Ionics 176 (2005) 2589 – 2598 www.elsevier.com/locate/ssi

Solid electrolytes and electrical interconnects for oxygen delivery devices Raymond A. Cutler a,*, D. Laurence Meixner b, Brett T. Henderson a, Kent N. Hutchings a, Dale M. Taylor a, Merrill A. Wilson a b

a Ceramatec, Inc., 2425 South 900 West, Salt Lake City, UT 84119, United States Air Products and Chemicals, Inc., 2425 South 900 West, Salt Lake City, Utah 84119, United States

Received 20 July 2004; received in revised form 3 August 2005; accepted 8 August 2005

Abstract High-purity oxygen can be produced using solid electrolytes, which allow oxygen migration under the influence of an externally applied electrical field, as governed by the Nernst relationship. Planar solid electrolyte oxygen separation (SEOS) technology uses lanthanide doped ceria as the electrolyte and electrically conductive lanthanum alkaline-earth manganites as the electrical interconnect between electrolyte plates. An inexpensive method for making electrolytes and interconnects is required to make this technology commercially viable. Ceria powder was mixed with Y, Gd, or Sm to achieve high ionic conductivity and a variety of sintering aids were explored. Titanium at a dopant level of less than 2 mol.% increases the sintering activity of Ce1
x
y Lnx Tiy O2
d electrolytes. A reactive sintering approach permits the use of inexpensive ceria, lanthanide, and titania powders to produce a solid solution in-situ. Biaxial stress testing of as-fired electrolytes was used to evaluate the sintered strength as a function of grain size. The effect of grain size on ionic conductivity is reported. The interconnect should have minimal ionic conductivity, low electrical resistivity, acceptable strength, and be expansion-matched to minimize thermal stresses upon cooling. Lanthanum alkaline-earth manganites ceramics satisfy these requirements by careful tailoring of their chemistry. The strength and modulus are significantly improved by substituting Ca2+ for Sr2+ in the structure. The substitution of Ca2+ for Sr2+ lowers the sintering temperature without compromising electrical conductivity. Thermal expansion was matched to the electrolyte at La0.4Ca0.6MnO3. Sintering temperatures below 1400 -C produce high-density interconnects with acceptable properties for operation in SEOS devices. D 2005 Elsevier B.V. All rights reserved. PACS: 60; 70; 81 Keywords: Solid electrolyte; Electrical interconnect; Oxygen delivery; Sintering; Ionic conductivity

1. Introduction Electrochemical devices for oxygen delivery using either zirconia or ceria as solid electrolytes have been developed in a number of geometries including tubular and planar designs [1 – 5]. The requirements for solid electrolyte oxygen separation (SEOS) devices differ from solid oxide fuel cells (SOFC) as has been discussed previously [4]. The fact that SEOS devices operate in air allows lanthanum strontium manganites (La1
x Srx MnO3Td or LSM), an early cathode

* Corresponding author. Tel.: +1 801 978 2126; fax: +1 801954 2008. E-mail address: [email protected] (R.A. Cutler). 0167-2738/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.ssi.2005.08.006

material for SOFC materials, to be used for the interconnect in SEOS systems. Early SEOS systems used yttria stabilized zirconia (YSZ) as the solid electrolyte where the LSM was expansion-matched to YSZ when x = 0.1. Due to the higher ionic conductivity of Gd or Sm-doped ceria [6– 8] as compared with YSZ, SEOS devices were designed with expansion-matched LSM, where x = 0.5 due to the higher thermal expansion of the ceria electrolyte. As x increases from 0.1 to 0.5 in LSM, strength and modulus decrease leading to some interesting plasticity in the interconnect [9,10]. One method for strengthening the interconnect is to make an undoped CeO2 – LSM composite [11]. A better approach is to replace Sr2+ with Ca2+, making LCM (La1
x Cax MnO3) so that the interconnect is expansion

2590

R.A. Cutler et al. / Solid State Ionics 176 (2005) 2589 – 2598

matched to the ceria electrolyte, has excellent electrical conductivity, and is much stronger than LSM [4,9]. In order for SEOS devices to be produced economically, it is important that the electrolyte and the interconnect be fabricated inexpensively. Ceria is difficult to densify when using powders with surface areas in the 2– 10 m2/g range without sintering additives [12]. Baumard, et al. [13] showed that small (0.1 to 3.0 wt. %) additions of Nb2O5 or TiO2 enhanced sintering. They hypothesized that Nb5+ substitutes for Ce4+ and compensates for Ce3+ formation, leading to disorder in the cationic sublattice, thereby affecting the cation mobility. In the case of TiO2, they speculated that CeO2 and TiO2 could react to form a pseudobinary compound CeTi2O6 that melts at temperatures in the range of 1360 – 1400 -C. Since densification in their experiments was complete by 1200 -C, this suggested that Ti+ 4 additions might play a different role. Chen and Chen [14] studied the effect of Mg2+, Ca2+, Sr2+, Sc3+, Yb3+, Y3+, Gd3+, La3+, Ti4+, and Nb5+ on the sintering of undoped ceria. Their results are in conflict with the previous results of Baumard, in that Nb2O5 additions did not enhance sintering compared to pure ceria at temperatures in excess of 1300 -C. Furthermore, while TiO2 additions were effective in promoting sintering, they were not as effective as Mg2+, and did not promote sintering at 1.0 mol. % additions. Severely undersized dopants (Mg2+, Sc3+, and Ti4+), compared to Ce4+, enhanced grain boundary mobility at 0.1 mol. % levels, but only Sc3+ had higher mobility at 1.0 mol. % additions. This decrease in cation mobility was explained based on solute drag, a well-know phenomenon in ceria [15]. More recently, Kleinlogel, et al. [16 – 18] used transition metal oxides to promote low-temperature sintering of gadolinia-doped ceria. Zhang, et al. [19] showed that cobalt oxide could be used to lower the sintering temperature of conventional powders. Small additions (< 2 mol. %) of transition metals enhanced sintering but had little effect on ionic conductivity at 700– 800 -C [16 –20]. The purpose of this paper is to report on inexpensive methods for making both electrolyte and interconnects for oxygen delivery devices. Ti4+ was used as the primary sintering aid in an in-situ reaction process which involved mixing inexpensive grades of CeO2, Sm2O3 and TiO2 to form Ce1
x
y Smx Tiy O2
d electrolytes tapes. The role of ceria grain size on the strength and ionic conductivity of the electrolyte is discussed. Furthermore, La2O3, CaCO3, and Mn3O4 were calcined to form a La0.4Ca0.6MnO3 interconnect. Properties of these novel LCM interconnects are compared with those of LSM.

prepared from oxide powders (purity was greater than 99.9 %). CeO2 (Rhone – Poulenc, Opaline ceramic grade) powder with a surface area of 5 – 6 m2/g was used for all compositions. Two hundred grams of each powder composition was dispersed in 70 grams of ethyl alcohol using 0.5 wt. % dispersant. The slurries were vibratory-milled inside 250 ml high-density polyethylene (HDPE) jars for 24 h with 1 kg high-purity Y – TZP (ZrO2 (3 mol. % Y2O3)) media. Two weight percent of a polyoxaline binder (Polymer Chemistry Innovations, Aquazol 50) was added to the slurries and the powders were stir-dried before screening through an 80 mesh screen. Bar-shaped samples were pressed uniaxially at 35 MPa and isostatically at 200 MPa, and heated in air on an MgO setter to temperatures ranging between 1300 and 1500 -C, with a 2-h hold at temperature. 2.1.2. In-situ tape preparation Lanthanide-doped ceria compositions (Ce1
x
y
z Lnx Tiy AEz O2
d , where Ln was either Sm, Gd, or Y, and AE was either Mg, Ca, or Sr) using Ti4+ in some cases alkaline earth elements as sintering aids were prepared by tape casting. Raw materials used were CeO2 (Rhone – Poulenc, Opaline ceramic grade), Sm2O3 (Molycorp grade 5810), Gd2O3 (Molycorp grade 5790), Y2O3 (Molycorp grade 5600), TiO2 (Degussa grade P-25), MgO (Baker chromatographic grade), CaCO3 (Baker grade 1294), and SrCO3 (Solvay grade SL 300). Appropriate powders were added to a 1-l HDPE jar with 1.5 kg Y – TZP media. The ceramic powders were mixed for 16 h in a toluene– ethanol (4 : 1 ratio based on mass) at a solids content of 25 vol. % using a dispersant. Binder (polyvinyl alcohol, Solutia grade B-79) and plasticizer (butyl benzyl phthalate, Monsanto grade S-160) were added at a 2.5 : 1 ratio (based on mass) predissolved in toluene –ethanol. The tape was cast with a doctor blade on polyester so that the dried thickness was approximately 300 Am. After evaporation of the solvents, the solids content of the tape was approximately 50 vol. %. The tape was sintered on porous SmAlO3 setters by heating to temperatures ranging between 1400 and 1550 -C for two hours in air.

2. Experimental

2.1.3. Fabrication of electrolytes Ce0.84Sm0.15Ti0.01O1.925 tape was scaled-up using the same concept of starting with inexpensive raw materials having a combined surface area of 4 – 6 m2/g. The slurry was cast on a production tape caster, sheared to the desired dimensions, and fired at 1500 -C for 4 h in air so that the sintered electrolyte had a nominal thickness of 250 Am. The effect of grain size on strength and ionic conductivity was determined using samples sintered at temperatures between 1500 and 1550 -C for times ranging from 3 to 12 hours.

2.1. Electrolyte preparation

2.2. Interconnect preparation

2.1.1. Screening different sintering aids Samaria-doped ceria compositions (Ce0.895Sm0.1M0.005 O2
d , where M = Al, Ce, Co, Mg, Nb, Ti, or W) were

Compositions of the general formula (La1
x
yCaxSry)A (Mn1
zCoz)BO3
d, where A/B represents the molar ratio of A-site (La, Sr, and Ca) and B-site (Mn and Co) cations, were

R.A. Cutler et al. / Solid State Ionics 176 (2005) 2589 – 2598

2.3. Sample characterization Densities of the sintered samples were determined using Archimedes’ method in water, with vacuum-induced infiltration to determine open porosity. Thinner electrolyte samples were measured with a helium pycnometer. Theoretical densities were calculated based on x-ray diffraction (XRD) lattice parameter measurements using Cu Ka radiation. Samples were fractured as bars in four-point bending or as disks in biaxial flexure using universal testing machines (Instron model 1123 or 5566). The loading rate was 0.5 mm/ min. In flexure tests, ground bars (approximately 3 mm  4 mm  50 mm) were chamfered and tested with a 40 mm support span and a 20 mm loading span. Elastic modulus measurements were made by attaching strain gages to a barshaped sample and loading-unloading the elastic-plastic region of the stress –strain curve in four-point bending. Biaxial strength tests were made using either the ball-onthree ball or hydraulic-burst techniques. For the ball-onthree ball method, sintered tape cast disks were tested supported on three 3.2 mm support balls, evenly spaced on a 12.7 mm diameter ring. A free-floating 3.2 mm diameter ball was used to apply the biaxial load at the center of the disk, which was centered over the three support balls. The biaxial strength is given by [21]:   2 
a   1
y  3F ð1 þ yÞ b2 a rf ¼ 1 þ 2ln þ 1
2 2 4pt b 1þy 2a R2 ð1Þ where F is the applied force, o is Poisson’s ratio, t is disk thickness, a is the radius of the circle of support points, b is

the radius of the region of ‘‘uniform’’ loading at the center, and R is the disk radius. The value of o was taken as 0.3 and b was taken to be approximately t/3 based on the work of Shetty et al. [22]. The hydraulic burst technique was based on the previous work of Shetty, et al. [23]. The disks were placed on a neoprene membrane that sealed the top of a reservoir filled with hydraulic fluid. The specimen was clamped by an upper fixture that contained an o-ring, known as the support ring. Pressure was applied to the hydraulic fluid which transmitted a uniform pressure to the specimens via the neoprene bladder. The region of highest stress occurred in the center of the disks. The strength of the disks, jf , was calculated from the maximum pressure, P, in the reservoir at failure using the following formula [24], (  2 3Pr21 r2 rf ¼ 2 2ð1
yÞ þ ð1 þ 3yÞ 8t r1

 2  ) r2 r2 ð 3 þ yÞ
4ð1 þ yÞ ln ð2Þ þP 4 ð 1
yÞ r1 r1 where r 1 is the radius of the support ring and r 2 is the specimen radius. Electrolyte disks with nominal values of t, r 1 , and r 2 of 250 Am, 12.5 mm, and 17 mm, respectively, were tested. Twelve to twenty-five samples were tested for each grain size. To quantify the plastic deformation of different samples at room temperature, sintered bars were loaded in three-point bending at room temperature and their permanent deformation was measured after the load was relaxed [10]. The degree of deformation, y, about the center point was defined, taking into account possible tilt of the sample, as [10]: z1 þ z2 d¼
zcent ð3Þ 2 where z 1 and z 2 represent the height 5 mm from the ends of the bar, and z cent denotes the height in the center of the bar. A 100

Co 95

% Theoretical Density

prepared. The raw materials used were: La2O3 (PIDC 99.999%), CaCO3 (GE grade 111-030-026), SrCO3 (Solvay grade SL 300), Mn3O4 (Chemetals H80 PF), and Co3O4 (OMG/APEX LS). Small powder batches (100 grams total) were prepared in 125 ml HDPE jars by vibratory milling with Y-TZP media at a ball-to-charge ratio of 3.5. Anhydrous methanol (40 grams) was used for each formulation and the milling time was 24 hours. The milled powders were dried and calcined on high-purity (99.8%) alumina plates at 1150 -C for 10 hours. The calcined powders were vibratory milled an additional 72 hours using the same procedure. The milled slurries were lubricated with 3 wt. % polyoxoline binder before pressing uniaxially into bars or pellets as described above. The pressed samples were sintered in air at temperatures ranging between 1100 and 1425 -C, with a four-hour hold at the peak temperature. Interconnects were fabricated by scaling up the composition La0.4Ca0.6Mn1.02O3 T d to 45 kg batches, which were calcined as above. The calcined powder was vibratorymilled in water and to a surface area of 4 m2/g and spray dried before pressing nominal 100 mm  100 mm interconnects uniaxially at 200 T. The parts were sintered for four hours at 1300 -C.

2591

Pr

Ceria 90

Al

Mg

Ti

85

Nb Ce

80

Ceria Al Ce Co Mg Nb Pr Ti W

75

W

70 65 1250

1300

1350

1400

1450

1500

1550

1600

o

Sintering Temperature ( C) Fig. 1. Densification characteristics of various Ce0.895Sm0.1M0.005O3
d compositions (M is Al, Ce, Co, Mg, Nb, Pr, or W) compared with CeO2.

2592

R.A. Cutler et al. / Solid State Ionics 176 (2005) 2589 – 2598

7.5

Table 1 Area Specific Resistance for Electroded 250 Am Thick Ce0.795Ln0.2 Ti0.005O1.9 at 750 -C

o

1500 C

Density (g/cc)

7

Ln

Density (g/cc)

Area Specific Resistance (V-cm2)

1450oC

Sm Gd Y

7.09 7.24 6.68

0.50 0.53 0.57

1400oC

conductivity measurements were made using a four-point probe with samples heated in air. Three measurements were made at currents between 0.3 and 1 A, for temperatures ranging between room temperature and 800 -C. Phases present were determined at room temperature using x-ray diffraction. A scanning electron microscope was used in both secondary and backscatter modes to examine microstructures. Grain size was determined by the linear intercept method [25] on polished microstructures after thermally etching at 1400 -C for 30 minutes, with each value reported being the average of 350– 475 grains.

6.5

6

5.5

5 0

0.5

1

1.5

2

Sintering Aid (Mol. %) Fig. 2. Effect of increasing concentration of sintering aids in Ce0.85
x Sm0.15(Ti0.5Mg0.5)xO1.925
x/2, where x varies from 0 to 0.02, on densification at three different temperatures.

value of y > 0 implies that the ends of the sample are higher than the center. This method was developed to quantify the surprising mechanical behavior of certain LSM materials and has been described previously [10]. A La1
xSrxCoO3
d electrode was screen printed onto both sides of the electrolyte and the fired electrodes were coated with a thick-film Ag current collector, with Ag leads extending out of the furnace. Two-probe electrolyte resistivity was calculated from the physical dimensions of the active cell (nominal dimensions of 250 Am by 2 cm by 2 cm) and the ohmic resistance of the cell. An impedance/ gain-phase analyzer (Solartron SI1260) with an electrochemical interface (Solartron SI1287) was used for impedance measurements. The impedance was measured over a range of 0.1 Hz to 20 kHz with an AC amplitude of 10 mV. An average of three cells in three different furnaces were used for resistivity measurements. Direct current electrical

3. Results and discussion 3.1. Electrolyte development The sintering behavior of various samaria-doped ceria compositions (Ce0.895Sm0.1M0.005O2
y, where M = Al, Ce, Co, Mg, Nb, Pr, Ti, or W) is compared to CeO2 in Fig. 1. Xray diffraction showed all samples to have the cubic fluorite lattice, as expected. Nb5+, Ti4+, and Mg2+ were chosen due to their ability to aid the sintering of undoped CeO2 [13,14]. Co2+ is a known sintering aid for ceria – lanthanide solid solutions [16 – 20]. W6+ was chosen since it is donor dopant which would work opposite of the acceptor dopants. Pr4+ was chosen due to its slightly larger ionic size than the other dopants. Al3+ was included for its potential as a grain 1

7.5 o

1500 C

0.8 Ohmic ASR (Ω-cm2)

Density (g/cc)

7 o

1450 C

6.5

6

o

1400 C

0.6

0.4

0.2 5.5 0 0

5 0

0.2

0.4

0.6

0.8

1

100

200

300

400

500

Electrolyte Thickness (µm)

Fraction Ti in One Mol. % Sintering Aid Fig. 3. Effect of mole fraction Ti sintering aid in Ce 0.84 Sm0.15 Ti0.01
xMgxO1.925
x on densification at three different temperatures.

Fig. 4. Effect of electrolyte thickness on the ohmic area specific resistance at 750 -C for Ce0.849Sm0.12Gd0.03Pr0.001O1.925, sintered at 1550 -C for two hours in contact with an MgO setter.

R.A. Cutler et al. / Solid State Ionics 176 (2005) 2589 – 2598

2 Ti-Mg σ o=291 MPa m=3.2

Ln Ln (1/(1-F))

1 0 -1

Ti σ o=530 MPa m=3.4

-2 -3 -4 4.5

5

5.5

6

6.5

7

Ln Strength (MPa) Fig. 5. Weibull plots showing ball-on-three-ball strength distributions comparing Ce 0.84 Gd 0.15 Ti 0.0067 Mg 0.0033 O 1.922 with Ce 0.84 Gd 0.15 Ti0.01O1.925. Characteristic stress (jo) and Weibull modulus (m) are given on plot.

growth inhibitor since it readily reacts with Sm to form SmAlO3. All materials had similar green densities of approximately 53% of theoretical after binder removal. Cobalt was the only additive which increased densification relative to undoped CeO2 at temperatures below 1400 -C, and was oversintered by 1500 -C. The introduction of Sm2O3 in CeO2 to form Ce0.9Sm0.1O1.95 by a reaction sintering approach slows the densification, as is easily observable in Fig. 1. Ti4+ clearly aided the sintering of samaria-doped ceria at all temperatures investigated and allowed near-theoretical density to be achieved at 1500 -C using this reaction sintering approach. Based on the work of

2593

Baumard, et al. [13] one might have predicted Nb5+ and Ti4+ to behave similarly, contrary to observation. Based on the work of Chen and Chen [14], again for undoped ceria, one would have expected Mg2+ and Nb5+ to be better sintering aids than Ti4+. Fig. 1 shows that this is not the case. Samples doped with tungsten showed the highest open porosities at temperatures below 1450 -C. The role of sintering aids in ceria solid solutions is poorly understood at present, but it is clearly different than in pure ceria. Titanium and magnesium were selected for further study based on these screening experiments. Fig. 2 shows the effect of equimolar ratios of Ti and Mg as sintering aids in electrolyte tape of thickness 250 Am for compositions of Ce0.85
xSm0.15(Ti0.5Mg0.5)xO1.925
x/2, where x varies from 0 to 0.02. Based on the work of Chen and Chen [14,15], one could expect slower diffusion for both Mg and Ti, as their concentrations increase, in accordance with solute drag. Fig. 2 shows that this is clearly different in the case of ceria solid solutions, where increasing the amount of sintering aid increases densification. The problem with adding too much sintering aid is that one exceeds solubility limits and end up precipitating secondary phases, which can lead to blocking behavior at the surface of the electrolyte if the phases which precipitate are not ionically conducting. The solubility of Ti4+ in CeO2 is less than that of Mg2+, which is reported as two mol. % at 1600 -C [6]. Ce0.845Sm0.15Ti0.0025AE0.25O1.93, where AE is Mg2+, Ca2+ or Sr2+, electrolytes were sintered at various temperatures. Under sintering conditions where Mg2+ reached greater than 99% of theoretical, Ca2+ was at 96.3% of theoretical and Sr2+ was at 95.7% of theoretical density. Undersized dopants aid in densification as taught by Chen and Chen [14,15].

Fig. 6. Polished and etched microstructures of Ce0.84Sm0.15Ti0.01O1.925. Markers are 5 Am.

R.A. Cutler et al. / Solid State Ionics 176 (2005) 2589 – 2598

d 2
do2 ¼ 2Mcðt
to Þ

ð4Þ

where d o is the initial grain size at time t o . Taking d o as 3.5 Am at 1550 -C when t o = 0 hr, and g as 1 J/m2 [29], the data in Fig. 7(b) were used to calculate M as 1.1 10
15 m3/Ns for Ce0.84Sm0.15Ti0.01O1.925. This value is an order of magnitude lower than the grain boundary mobility of Ce0.99Ti0.01O2 [16]. The grain boundary mobilities of ceramics cover a wide range and Ce0.84Sm0.15Ti0.01O1.925 is between Y2O3 and Ce-TZP [29].

(a) 7

Grain Size (µm)

Ti4+ is a more effective sintering aid in CeO2 solid solutions than Mg2+ as shown in Fig. 3. The substitution of Gd3+ or Y3+ for Sm3+ does not change this result and Ti4+ is effective for sintering a wide variety of lanthanide-doped ceria compositions. The area specific resistance (ASR) values for all three of the highest conductivity lanthanide [6 –8] measured at 750 -C were comparable to those of ceria electrolytes made by processing expensive powders (see Table 1). As expected, the ohmic electrolyte resistance is the major contributor to the cell resistance at the electrolyte thickness range investigated. The electrolyte used to obtain the ASR data shown in Fig. 4 initially contained no sintering aids. However, by sintering in contact with MgO setters at 1550 -C, Mg2+ and impurities from the setter diffused into the ceria electrolyte. This method for introducing sintering aids lacks control, even though Fig. 4 demonstrates that high ionic conductivity can be obtained in this manner. Lanthanide aluminates (LnAlO3, Ln3Al5O12, or Ln2Al22O36, where Ln is Y, La, Sm, or Gd depending on the equilibrium phase stability of Ln) make inexpensive setters and limit diffusion when using the reactive sintering approach. SmAlO3, for example, is an excellent setter for reactively sintering a wide range of compositions at temperatures up to 1550 -C. There is no advantage from a strength perspective in partial substitution of an alkaline-earth ion for Ti as shown by the Weibull plots in Fig. 5. A clear indication of a threshold stress is evident from these data. At concentrations of one mole % Ti4+ or less, there is no indication of any secondary phase formation based on energy dispersive spectroscopy. Fig. 6 and Table 2 show grain sizes as a function of temperature and time at temperature. Fig. 7(a) shows that grain growth of Ce0.84Sm0.15Ti0.01O1.925 is thermally activated (Ea = 342 kJ/mol), as expected. Activation energies for grain growth of pure ceria vary widely from values of 314 kJ/ mol [26] to 731 kJ/mol [27]. A value of 433 kJ/mol for the grain growth of Ce0.8Sm0.2O1.9 was reported by Okawa, et al. [28]. The activation energy of Ce0.84Sm0.15Ti0.01O1.925 is relatively low for ceria-based ceramics (Fig. 7). The grain growth of Ce0.84Sm0.15Ti0.01O1.925 followed a parabolic growth law consistent with undoped ceria as reported by Chen [29]. If grain boundary mobility (M) and energy (g) are independent of grain size, the grain size, d, at time t is given by [29]

6

Ea=342 kJ/mol

5

4

3 0.545

0.55

0.555 1000/T (K

0.56

0.565

-1)

(b) 100 80 d2-do2(µm2)

2594

60 40 20 0 0

2

4

6

8

10

12

Time (hr) Fig. 7. (a) Arrhenius plot of Ce0.84Sm0.15Ti0.01O1.925 grain size as a function of inverse absolute temperature. Activation energy is 342 kJ/mol. (b) Grain growth of Ce0.84Sm0.15Ti0.01O1.925 at 1550 -C as a function of annealing time. Grain sizes are do and d at times 0 and t, respectively. Slope of line is 8.2.

Ionic conductivity of ceria solid solutions are sensitive to the choice of the dopant, the concentration of the dopant, the grain size, and the test temperature [30]. Fig. 8 shows conductivity data for different grain sizes of Ce0.84Sm0.15 Ti0.01O1.925. Ionic conductivities were strongly dependent on the test temperature of the electrolyte cell, but were only weakly dependent on the grain size over the temperature range (500 – 750 -C) investigated. The activation energies, which ranged between 69 and 72 kJ/mol, agree well with a value of 69 kJ/mol based on the work of Hong, et al. [30]. There is little influence of grain boundary resistivity for the grain sizes investigated, especially at operating temperatures for oxygen generation. The strength of the electrolyte is important in an oxygen generation device since it is much thinner than the interconnect [4] and can limit the strength of the device. Hydraulic-burst biaxial strength measurements [23,24] are well-suited for determining strength since they can be made on as-fired surfaces [31] and put a larger volume of material

R.A. Cutler et al. / Solid State Ionics 176 (2005) 2589 – 2598

4.5

100

2595

1425oC

4 % Theoretical Density

90

σ T (S/cm-T)

3.5 3 2.5 1500 oC/3 hr

2

1525 oC/3 hr 1550 oC/6 hr

1.5 1 0.95

1325oC 80

70 1225oC 60 1125oC

o

1550 C/12 hr

50 1

1.05

1.1

1.15

1000/T

(K-1)

1.2

1.25

0

1.3

0.2

0.4

0.6

0.8

1

Mole Fraction Ca/(Ca+Sr) on A-site

Fig. 8. Temperature dependence of ionic conductivities of Ce0.84Sm0.15 Ti0.01O1.925 for grain sizes ranging between 3 and 11 Am. Activation energies are approximately 72 kJ/mol.

Fig. 9. Effect of Ca content on the densification behavior of La0.4Ca0.6
xSrxMn1.02O3
d sintered at temperatures between 1125 and 1425 -C for four hours.

under stress as compared to the ball-on-three ball test. The strength data, shown in Table 2, show little dependence on grain size since the strength distributions overlap within one standard deviation. Taking the fracture toughness of ceria solid solutions as 1.5 MPa¾m [32,33] and assuming fracture initiation from an edge crack, the critical defect sizes would be 8.4 and 11.8 Am for samples with strengths of 220 and 260 MPa, respectively. These values are within a factor of two of the mean grain sizes. A mean grain size of 4– 5 Am gives acceptable strength and ionic conductivity for electrolytes used in oxygen generation systems. A reaction-sintered Ce0.84Sm0.15Ti0.01O1.925 electrolyte using inexpensive raw materials can be made with a density greater than 99% of theoretical, a grain size of approximately 5 Am, biaxial strength of 250 MPa, and an ionic conductivity of 5.6  10
2 S/cm at 750 -C, comparable with similar materials made using more expensive coprecipitated starting powders [34].

upon cooling. This is easily accomplished in LSM by adjusting the La3+ to Sr2+ ratio, with thermal expansion decreasing as La3+ increases. Zirconia, for example, is expansion-matched with La0.9Sr0.1MnO3 T d, whereas ceria is expansion-matched with La0.5Sr0.5MnO3
d [11]. At equimolar lanthanum and strontium the LSM displays unusual plastic deformation, which is dependent on oxygen stoichiometry [10]. While plastic deformation is desirable when accompanied by robust mechanical properties, it is not desirable when the material has low strength. Due to prior work with LCM as a giant magnetoresistive material [35], there are data indicating that these LCM can be sintered at lower temperatures than LSM [36,37]. This is clearly shown in Fig. 9, where La0.4Ca0.6
xSrxMn1.02O3
d is sintered at various temperatures. The higher densification rate of LCM compared with LSM is clearly evident from this plot, which shows a 100– 200 -C temperature differential for the same sintered density in the range of 1125 to 1325 -C. At 1325 -C the LCM is nearly theoretically dense with no open porosity, while the LSM has nearly 30% open porosity. Sintering above 1400-C allows both LSM and LCM to be fired to near theoretical density. The data in Fig. 9 also show that 5% of the Ca2+ can be substituted with Sr2+ without any

3.2. Interconnect development The interconnect must be expansion-matched to the electrolyte to avoid the introduction of residual stresses Table 2 Effect of Grain Size on Strength and Ionic Resistivity of Ce0.84Sm0.15Ti0.01O1.925 Sintering conditions

Density

Grain Size

Resistivity

Biaxial strength (MPa)a

Temp.(-C)

Time(hr)

(g/cc)

(Am)

(V-cm)

Meanb

Char.c

md

1500 1525 1550 1550 1550

3 3 3 6 12

6.98 T 0.01 7.09 T 0.01 7.09 T 0.01 7.08 T 0.01 7.08 T 0.01

3.6 T 0.1 4.7 T 0.5 6.8 T 0.6 7.4 T 0.4 10.7 T 0.2

19.4 T 0.3 18.0 T 0.5 17.8 T 0.4 17.8 T 0.3 17.5 T 0.3

261 T 75 254 T 57 222 T 66 238 T 57 220 T 59

287 276 247 259 242

4.2 5.3 3.6 4.8 4.5

a b c d

Hydraulic burst technique. Mean strength with T one standard deviation. Characteristic strength. Weibull modulus.

R.A. Cutler et al. / Solid State Ionics 176 (2005) 2589 – 2598

% Theoretical Density

100 1425oC

90 1325oC o

1225 C

80

70

o

1125 C

(a) 35 30 LCM E=165 GPa

25 Stress (MPa)

effect on densification. Further strontium substitutions lower the densification kinetics. LCM is expansion-matched with the Ce0.84Sm0.15 Ti0.01O1.925 electrolyte at a La3+/Ca2+molar ratio of 0.67. Fig. 10 shows the densification sensitivity with respect to A/ B ratio for this composition. As expected, the densification is very sensitive to the A/B ratio, with stoichiometric or Bsite rich compositions sintering more readily than A-site rich compositions [9,37,38]. The rate-limiting step in the densification process is the La3+ mobility and creating Asite vacancies speeds up the sintering rate substantially. While sintering aids may be added to LCM compositions, adequate sinterability occurs without any additional additives. Keeping the composition slightly A-site deficient allows for densification of interconnects to greater than 97 % of theoretical density at 1300 -C. The substitution of Ca2+ for Sr2+ in lanthanum manganites makes a dramatic effect on the room-temperature plasticity as shown in Fig. 11. The LSM (La0.5Sr0.5Mn1.02 O3
d) shows elastic-plastic deformation at low loads, whereas LCM (La0.4Ca0.6Mn1.02O3) displays the expected elastic behavior (see Fig. 11(a)). The degree of permanent plastic deformation after unloading, y, is a function of applied stress for LSM, as shown in Fig. 11(b), whereas no plasticity is measured for LCM. This increase in stiffness is accompanied by a flexural strength of LCM which is more than double that of LSM, as shown in Table 3. The strength and stiffness of LCM is typical of many perovskites and it is the plasticity of LSM that is unusual [10]. Nonstoichiometry in lanthanum manganites is well recognized [39 –44] but both LSM and LCM are stoichiometric over a wide partial pressure range [44]. While it was not possible to fully explain the unusual plasticity of LSM, it was very apparent that it was related to oxygen nonstoichiometry [10] and it is therefore not surprising that LCM behaves as an elastic

20 15 10 5

LSM E=29 GPa

0 0

0.0002 0.0004 0.0006 0.0008 0.001

0.0012

Strain

(b) 160 Deformation Parameter, δ (µm)

2596

140 LSM

120 100 80 60 40 20

LCM 0 0

5

10

15

20

25

30

35

Stress (MPa) Fig. 11. Room-temperature flexural loading of LSM (La0.5Sr0.5Mn1.02O3
d) and LCM (La0.4Ca0.6Mn1.02O3
d) showing plastic and elastic behaviors, respectively. (a) Stress – strain behavior, (b) permanent deformation as measured by y parameter.

ceramic with properties typical of many low-toughness ceramics. The electrical conductivity of many perovskites, including LSM and LCM, follow a small polaron hopping mechanism such that electrical conductivity times absolute temperature is an exponential function of temperature, as shown in Fig. 12. There is little dependence of electrical conductivity on temperature, as activation energies are 5.8 kJ/mol. The electrical conductivities of the two materials are similar over a wide temperature range. This is consistent with the work of Mackor, et al. [36] who reported values of 230 and 240 S/cm for electrical conductivities of

60

50

Table 3 Room-Temperature Flexural Strength of LSM and LCM

0.96

0.98

1

1.02

1.04

A/B Molar Ratio Fig. 10. Effect of A/B on the densification behavior of (La0.04Ca0.6)A (Mn)BO3
d sintered at temperatures between 1125 and 1425 -C for four hours.

Perovskite

Number of Strength (MPa) Weibull Samples Mean Characteristic Modulus Tested

La0.5Sr0.5MnCo0.04O3
d 23 22 La0.4Ca0.6MnO3
d

57 T 4 59 146 T 16 153

17 11

R.A. Cutler et al. / Solid State Ionics 176 (2005) 2589 – 2598

13

σeT (S/cm-T)

12.5

LSM LCM

12

11.5

11

10.5 0.5

1

1.5

2 2.5 1000/T (K-1)

3

3.5

Fig. 12. Temperature dependence of interconnect electrical conductivity of LSM and LCM. Activation energies are approximately 5.8 kJ/mol.

La0.45Sr0.45MnO3
d and La0.45Ca0.45MnO3, respectively, at 1000 -C. La0.4Ca0.6Mn1.02O3 had values of 175 and 320 S/cm at 25 and 750 -C, respectively. LCM is an excellent interconnect material for oxygen generation devices because it is expansion-matched to Ce0.84Sm0.15Ti0.01O1.925, has excellent dimensional stability due to its higher stiffness relative to LSM, has adequate strength, and has high electrical conductivity without ionic conductivity. For applications where plasticity is desired, Sr can be substituted for Ca on the A-site in order to tailor the degree of plasticity desired. LCM interconnects, produced by uniaxially pressing spray dried powder, can be sintered to greater than 97% of theoretical density at 1300 -C.

4. Conclusion A reaction-sintering approach was used to make a variety of ceria-based electrolytes using starting oxides with moderate surface areas. These inexpensive starting materials were densified at temperatures below 1550 -C in air on lanthanide aluminate setters. Ti4+ is an excellent sintering aid for ceria solid solutions because it allows densification to greater than 99% of theoretical density. Alkaline-earth oxide ions, particularly Mg2+, can be used in combination with Ti4+ to promote sintering but they offer no advantage when compared with Ti4+ as a sintering aid. Lanthanidedoped ceria, made by reaction sintering, shows resistivity below 20 V-cm and cell (electrolyte plus electrodes) areaspecific resistance below 0.3 V-cm2 at 750 -C for 100 Am thick electrolyte plates. These values are comparable to the electrolytes made using more expensive coprecipitated powders. A reaction-sintered Ce0.84Sm0.15Ti0.01O1.925 electrolyte using inexpensive raw materials can be made with a density greater than 99% of theoretical, a grain size of approximately 5 Am, biaxial strength of 250 MPa, and an ionic

2597

conductivity of 5.6  10
2 S/cm at 750 -C. Grain growth, which is parabolic and thermally activated (Ea = 342 kJ/ mol), decreases both the strength and the ionic conductivity slightly for grain sizes between 3 and 11 Am. The substitution of calcium for strontium in alkalineearth doped lanthanum manganites lowers their sintering temperature by 100 -C, and increases their stiffness and strength by a factor of two to three. LCM is an excellent interconnect material for oxygen generation devices because it is expansion-matched to Ce0.84Sm0.15Ti0.01O1.925, has minimal ionic conductivity, and electrical conductivity above 300 S/cm at operating temperatures for oxygen delivery devices. This material can be made inexpensively starting with oxides by calcination and spray drying prior to sintering at 1300 -C in air. Acknowledgments The technical assistance and fruitful technical discussions with Dr. Stuart Adler, Ms. Charla Brinkpeter, Dr. Michael F. Carolan, Mr. Marc Flinders, Mr. Jeff Kawola, Mr. Brian Kleinlein, Mr. Darin Ray and Dr. Robin E. Richards are gratefully acknowledged. The permission of Air Products and Chemicals and Ceramatec to publish this work is appreciated. This effort was supported in part by Air Force Research Laboratory, Biosciences and Protection Division, Brooks City-Base, San Antonio TX (U.S. Air Force Research and Development Contract No. F41624-00-C-6000).

References [1] [2] [3] [4]

[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19]

J. Weissbart, W. Smart, T. Wydeven, AMSE Paper, 1971. S.P.S. Badwal, D.A. Graham, Key Eng. Mater. 53-55 (1991) 53. J.W. Suitor, Solid State Ionics 52 (1992) 277. D.L. Meixner, D.D. Brengel, B.T. Henderson, J.M. Abrado, M.A. Wilson, D.M. Taylor, R.A. Cutler, J. Electrochem. Soc. 149 (2002) D132. F.T. Ciacchi, S.P.S. Badwal, V. Zelizko, Solid State Ionics 152-153 (2002) 763. T.H. Etsell, S.N. Flengas, Chem. Rev. 70 (1970) 339. K. Eguchi, T. Setoguchi, T. Inoue, H. Arai, Solid State Ionics 52 (1992) 165. H. Inaba, H. Tagawa, Solid State Ionics 83 (1996) 1. D.L. Meixner, R.A. Cutler, Solid State Ionics 146 (2002) 273. D.L. Meixner, R.A. Cutler, Solid State Ionics 146 (2002) 285. R.A. Cutler, D.L. Meixner, Solid State Ionics 159 (2003) 9. I. Riess, D. Braunshtein, D.S. Tannhauser, J. Am. Ceram. Soc. 64 (1981) 479. J.F. Baumard, A. Argoitia, C. Gault, J. Less-Common Met. 127 (1986) 125. P.L. Chen, I.W. Chen, J. Am. Ceram. Soc. 79 (7) (1996) 1793. P.L. Chen, I.W. Chen, J. Am. Ceram. Soc. 77 (9) (1994) 2289. C. Kleinlogel, L.J. Gaukler, Proc. Electrochem. Soc. 99-19 (1999) 225. C.M. Kleinlogel, L.J. Gaulker, J. Electroceram. 5 (3) (2000) 231. C. Kleinlogel, L.J. Gaukler, J. Electroceram., Adv. Mater. 13 (2001) 1081. T. Zhang, P. Hing, H. Huang, J. Kilner, J. Eur. Ceram. Soc. 22 (1) (2002) 27.

2598

R.A. Cutler et al. / Solid State Ionics 176 (2005) 2589 – 2598

[20] G.S. Lewis, A. Atkinson, B.C.H. Steele, J. Drennan, Solid State Ionics 152-153 (2002) 567. [21] A.F. Kirstein, R.M. Woolley, J. Res. Natl. Bur. Stand. 71C (1967) 1. [22] D.K. Shetty, A.R. Rosenfield, P. McGuire, G.K. Bansal, W.H. Duckworth, Am. Ceram. Soc. Bull. 59 (1980) 1193. [23] D.K. Shetty, A.R. Rosenfield, W.H. Duckworth, P.R. Held, J. Am. Ceram. Soc. 66 (1) (1983) 36. [24] A. Simpatico, W.R. Cannon, M.J. Matthewson, J. Am. Ceram. Soc. 82 (10) (1999) 2737. [25] E.E. Underwood, Quantitative Stereology, Addison-Wesley, Reading, MA, 1970. [26] S. Okikawa, S. Somiya, S. Saito, Yogyi Kyokaishi 79 (1971) 365. [27] T. Zhang, P. Hing, H. Huang, J. Kilner, Mat. Lett. 57 (2) (2002) 507. [28] Y. Okawa, T. Matsumoto, T. Doi, Y. Hirata, J. Mater. Res. 17 (9) (2002) 2266. [29] I.W. Chen, Interface Sci. 8 (2000) 147. [30] S.J. Hong, K. Mehta, A.V. Virkar, J. Electrochem. Soc. 145 (2) (1998) 638. [31] C.A. Lewinsohn, M. Wilson, D. Taylor, Proc. Electrochem. Soc. 2002-2026 (2003) 364. [32] J.E. Shemitt, H.W. Williams, M.J. Edrisinghe, J.R.G. Evans, B. Ralph, Scr. Mater. 36 (8) (1997) 929.

[33] T.S. Zhang, J. Ma, L.B. Kong, P. Hing, J.A. Kilner, Solid State Ionics 167 (2) (2004) 191. [34] S. Zha, C. Xia, G. Meng, J. Power Sources 115 (2003) 44. [35] C.N.R. Rao, A.K. Cheetham, R. Mahesh, Chem. Mater. 8 (1996) 2432. [36] A. Mackor, T.P.M. Koster, J.G. Kraajkamp, J. Gerretsen, Proc. Electrochem. Soc. 1991-?? (1991) 463. [37] J.W. Stevensen, T.R. Armstrong, W.J. Weber, Proc. Electrochem. Soc. 1995-1 (1995) 454. [38] J.A.M. van Roosmalen, E.H.P. Cordfunke, J.P.P. Huijsmans, Solid State Ionics 66 (1993) 285. [39] J. Mizusaki, N. Mori, H. Takai, Y. Yonemura, H. Minamiue, H. Tagaw, M. Dokiay, H. Inaba, K. Naraya, T. Sasamoto, T. Hashimoto, Solid State Ionics 129 (2000) 163. [40] K. Yasumoto, N. Mori, J. Mizusaki, H. Tagawa, M. Dokiya, J. Electrochem. Soc. 148 (1) (2001) A105. [41] K. Yasumoto, N. Mori, H. Tagawa, J. Mizusaki, M. Dokiya, J. Electrochem. Soc. 149 (5) (2002) A531. [42] K. Yasumoto, J. Mizusaki, H. Itoh, S. Wang, H. Tagawa, M. Dokiya, Proc. Electrochem. Soc. 2003-07 (2003) 458. [43] J. Nowotny, M. Rekas, J. Am. Ceram. Soc. 81 (1) (1998) 67. [44] L. Rørmark, K. Wiik, S. Stølen, T. Grande, J. Mater. Chem. 12 (4) (2002) 1058.